-
River Otters (Lontra canadensis) in Southcentral Alaska:
Distribution, Relative Abundance, and Minimum Population
Size
Based on Coastal Latrine Site Surveys
Final Report
July 30, 2009
Merav Ben-David University of Wyoming
Department of Zoology and Physiology 1000 E. University Ave.
Laramie, WY 82071 [email protected]
Howard Golden
Alaska Department of Fish and Game Division of Wildlife
Conservation
PO Box 25526 Juneau, Alaska 99802
[email protected]
RM-CESU Cooperative Agreement Number: H1200040001
Funding & Support Provided by: National Park Service,
Inventory & Monitoring Program
National Science Foundation US Forest Service, Chugach National
Forest
Oil Spill Recovery Institute University of Wyoming, NSF-EPSCoR
program
Alaska Department of Fish and Game
1
mailto:[email protected]:[email protected]
-
File Name: BenDavidM_2009_SWAN_RiverOtter_FinalRpt.doc
Recommended Citation: Ben-David M. and H. N. Golden. 2009. River
Otters (Lontra canadensis) in Southcentral Alaska: Distribution,
Relative Abundance, and Minimum Population Size Based on Coastal
Latrine Site Surveys. SWAN I&M program report, National Park
Service. Anchorage, AK. 43 pg. Topic(s): Inventory and Monitoring,
Biological Theme Keywords: Lontra Canadensis, river otter, density
estimation, mark-recapture, latrine, feces, DNA analysis, genetic
monitoring. Placename Keywords: Alaska, Southcentral Alaska, Kenai
Peninsula, Prince William Sound, Kenai Fjords National Park, Katmai
National Park and Preserve, Lake Clark National Park and Preserve,
Kodiak Island Archipelago, Upper Cook Inlet. Acronyms: EVOS Exxon
Valdez Oil Spill GIS Geographic Information System I&M
Inventory & Monitoring Program KATM Katmai National Park KEFJ
Kenai Fjords National Park and Preserve KOD Kodiak Island
Archipelago LACL Lake Clark Nationa Park and Preserve NPS National
Park Service PWS Prince William Sound SWAN Southwest Alaska
Inventory & Monitoring Network Initial Distribution: Kenai
Fjords National Park– 1 hardcopy, 1 electronic Southwest Alaska
Network – 1 hardcopy, 1 electronic Abstract
In the aftermath of the Exxon Valdez oil spill (EVOS), studies
of coastal river otters (Lontra canadensis) in PWS indicated they
are a keystone species for the land-margin ecosystem and a sentinel
species for monitoring levels of environmental contamination. In
2004-2007, we surveyed latrine sites and used DNA fingerprinting to
establish baseline information on the distribution, relative
abundance, and minimum number of river otters alive in KEFJ, KATM,
LACL, PWS, and KOD, Alaska. We also assessed connectivity and
geneflow among these otter populations and identified parameters
that may contribute to differences in abundance and distribution of
river otters in these areas. We surveyed approximately 2,000 km of
shoreline and sampled fresh feces from
2
-
641 active otter latrines. Each site was visited between 1 and 9
times during the sampling periods. Fecal DNA analyses revealed that
all populations were genetically distinct as expected from
isolation by distance models, although otters inhabiting KOD were
isolated from their mainland conspecifics in KATM despite the
relatively short distance between these 2 shorelines. Latrine
density (sites/km) varied from 0.20 to 1.90 and fecal deposition
rate (feces/site/day) ranged from 0.82 to 4.77. The naïve density
ranged from 0.07 known individuals per km to 0.68 otters per km.
Finally, we found no relation between latrine density and fecal
deposition rate and the minimum number of otters known alive.
Executive Summary This report summarizes the activities in calendar
years 2004 – 2009 of a multi-agency collaborative project. Surveys
in Kenai Fjords National Park (KEFJ), Katmai National Park and
Preserve (KATM), Lake Clark National Park and Preserve (LACL),
Prince William Sound (PWS) and the Kodiak Island Archipelago (KOD)
were conducted under separate study plans and funding, but utilize
the same methodologies to meet similar goals. All projects and
datasets are presented here for comparative purposes and to better
illustrate the regional scale of this effort. The goals of this
study were: : (1) to establish baseline information on the
distribution, relative abundance, and minimum number known alive of
river otters in KEFJ, KATM, LACL, PWS, and KOD using latrine site
surveys and DNA fingerprinting of fecal samples; (2) to conduct
formal mark recapture analysis for estimating numbers of river
otters in KEFJ and PWS from DNA fingerprinting of fecal samples;
(3) compare patterns of abundance and distribution of river otters
in KEFJ, KATM, LACL, PWS and KOD; (4) assess the relation between
formal population estimates and indices of abundance (distribution
and use of latrine sites and the deposition rates of scats); and
(5) identify parameters that may contribute to differences in
abundance and distribution of river otters among the 5 areas.
Survey of the coastline for latrine sites in KEFJ was conducted in
July 5-10, 2004 and June 25-29, 2005. In 2006, a survey of random
sites was conducted from June 20-24 in that location. In KATM,
latrine site survey was conducted in July 2-10, 2005, and in LACL
in July 25-29, 2006. In PWS, a sound-wide survey was conducted in
August 9-21, 2004 and a more localized survey in Herring Bay, Lower
Passage, and Eleanor Island from May 25 – August 15, 2006. In KOD
survey was conducted 13–25 June, 2007. The length of shoreline
surveyed at each location ranged from 60-945 km and the number of
occasions varied from 1 in KEFJ in 2004, KATM in 2005, and KOD in
2007, to 9 in PWS in 2006. In PWS 2004 a second occasion occurred
48 hours after the first; in KEFJ 2005 a second occasion occurred
24 hours after the first. In PWS 2006 the interval between
occasions was 5-7 days. Latrine density (sites/km) varied from 0.35
to 1.90 and fecal deposition rate (feces/site/day) ranged from 0.82
to 4.77. Between 22 – 65% of collected samples yielded DNA. Higher
yield occurred in PWS 2006 and KOD 2007. Our analyses indicate that
to reduce time, effort, and cost associated with amplifying
poor-quality samples,
3
-
observers should preferentially collect samples that contain
anal gland secretions. Also, any sample should be discarded if it
does not amplify after three PCRs with the most reliable primers,
and one or more of the following conditions apply: 1) it contains
parasites, 2) it contains remains of Cottidae or Pholidae, or 3) it
was collected when ambient temperatures exceeded 16˚C. There was no
observer bias in the identification of samples that would yield
DNA. Also, none of the habitat features that we measured could
explain genotyping success or failure. Thus, it will be possible to
train novice observers to identify otter latrines based on habitat
features as well as to collect high quality samples. Of the 2,553
fresh otter feces we collected, 603 yielded consensus genotypes at
7 or more loci (KEFJ – 120 samples; PWS – 374; KATM – 15; KOD –
94). These samples represented 422 unique individuals (KEFJ – 103
samples; PWS – 237; KATM – 12; KOD – 82). The probability of
obtaining an incorrect multilocus genotype after replication at all
eight microsatellite loci ranged from 0.0004 to 0.009 across all
populations. Allelic richness and heterozygosity were high in all
areas except KOD. The probability that two individuals drawn at
random from a given population share identical genotypes at all
loci was low (KEFJ – 1 in 15,948,963; PWS – 1 in 7,575,758; KATM –
1 in 333,333; KOD – 1 in 38,760). Except for PWS 2006, recapture
rate of individuals was low and most recaptured individuals were
detected on the same day, precluding the calculation of formal
abundance estimates. In PWS in 2006 we identified 58 individuals
that were encountered between two and eight times and additional 73
individuals that were observed only once yielding an estimate of
123 (± 29) river otters (or 1 otters per 1.18km of shoreline). Our
analyses revealed that a minimum of 6 occasions will be required
for obtaining unbiased and precise population estimates for coastal
river otters. Finally, an analysis combining the data from all
subpopulations of otters in this study and a companion study in
British Columbia indicated no relation between latrine density and
fecal deposition rate and the minimum number of otters known alive.
Thus, the status of river otter populations could not be assessed
accurately with indices such as latrine density and/or fecal
deposition rate. Our results also indicate that otters in
southcentral and southwest Alaska belong to genetically distinct
populations with isolation by distance as the main mechanism
leading to differentiation. Nonetheless, despite the relatively
short distance between the Kodiak Island Archipelago and the Alaska
Peninsula (approximately 50 km away), KOD animals appear to be as
isolated genetically from their mainland conspecifics as otters
inhabiting PWS are from those from British Columbia. Our results
also indicate that KATM and KOD otters likely differentiated from
one ancestral stock that inhabited the Pleistocene southwestern
shores of Alaska, and was isolated from other more easterly
populations by distance. In addition, although the straight line
distance between KEFJ and KATM is shorter (160 km) than that
between PWS and KATM (325 km), animals from the latter appeared to
genetically cluster with those from PWS. Also, estimated migration
rates between these 2 populations were slightly higher and the
pairwise FST value slightly lower than those between KEFJ and KATM.
Although geneflow between KEFJ and PWS was higher than among other
populations, none of the KEFJ individuals was mis-assigned to the
PWS-KATM cluster with probability greater than 0.5 and only 16
individuals from PWS (or 6.7%) were mis-assigned to the KEFJ
cluster with
4
-
probability greater than 0.7. Finally, our analyses indicate
that otters in KEFJ belong to two sub-populations, with one cluster
assigned to the coast between Resurrection Bay and Harris Bay and
the second found along the shores of McCarthy Fjord and Nuka Bay.
Thus, the high levels of population structuring among and within
southwestern and southcentral Alaska should be incorporated into
future management decisions of this species and its habitat. Our
inventory surveys of the coastline in KEFJ, KATM, and LACL,
highlighted the differences in habitat availability for river
otters. Much of the coastline in LACL consists of muddy tidal flats
that are selected against by river otters. In our survey of the
shoreline we found few latrines and none of the samples we
collected yielded viable DNA. Nonetheless, our results suggest that
this area may be an important corridor for geneflow among otter
populations in southcentral and southwest Alaska, and thus should
be afforded high levels of protection. Introduction In the
aftermath of the Exxon Valdez oil spill (EVOS), studies of coastal
river otters (Lontra canadensis) in Prince William Sound (PWS)
indicated they are a keystone species for the land-margin ecosystem
and a sentinel species for monitoring levels of environmental
contamination (Bowyer et al. 2003). River otters in coastal
environments of Alaska tend to select old-growth forest habitats
close to the shore, where their chief food items are marine
bottom-dwelling fishes (Larsen 1983; Bowyer et al. 1994). The
effects of oil contamination and logging on habitat use, movements,
and food habits of river otters indicate these animals are
sensitive to disturbance by humans (Bowyer et al. 2003). Because of
their role as keystone species for the land-margin, responses of
river otter to human disturbances, such as oil contamination,
logging, harvest, construction of dwellings, and heavy recreational
use, should be addressed. Also, river otters in coastal Alaska rely
on nearshore resources similar to sea otters (Enhydra lutris) and
seals (Phoca vitulina) and thus represent a reliable model for the
responses of nearshore predators to human disturbance in coastal
areas of Alaska. River otters have been identified as a vital sign
by the Kenai Fjords National Park (KEFJ), Katmai National Park and
Preserve (KATM), and Lake Clark National Park and Preserve (LACL)
monitoring programs. The effects of human disturbances, such as
logging, harvest, and construction of dwellings and other
disturbances such as oil contamination and heavy recreational use
on river otters within these park units should be evaluated. No
information on the abundance and distribution of otters in the 3
areas was available before the initiation of this effort. The goals
of this study were:
1. To establish baseline information on the distribution,
relative abundance, and minimum number known alive of river otters
in KEFJ, KATM and LACL using latrine site surveys and DNA
fingerprinting of fecal samples.
2. To conduct formal mark-recapture analysis for estimating
numbers of river otters in KEFJ and PWS from DNA fingerprinting of
fecal samples.
3. Compare patterns of abundance and distribution of river
otters in KEFJ, KATM, LACL, PWS, and KOD.
5
-
4. Assess the relation between formal population estimates and
indices of abundance (distribution and use of latrine sites and the
deposition rates of scats).
5. Identify parameters that may contribute to differences in
abundance and distribution of river otters among the 5 areas.
River otters inhabiting marine environments deposit marine
nutrients onto the land by marking specific locations along the
shoreline with feces, urine, and anal gland excretions (Bowyer et
al. 1995). Known as latrine sites, these areas are typically 10–50
m in radius and 25–700 m apart (ca. 160 latrines/ 100 km of
shoreline). The location of latrine sites along the coast is
dependent on several habitat variables related mainly to features
of the intertidal zone, such as tidal slope, rock size, and extent
of Laminaria sp. beds (Ben-David et al. 1996; 2005; Bowyer et al.
1995). Nutrient transports by river otters to terrestrial
landscapes can be significant. With densities ranging from 1 otter
per 2.7 km (Melquist and Hornocker 1983) to 1 otter per 1.3 km of
shoreline (Testa et al. 1994), 100 km of shoreline may receive from
536 to 1,112 kg of N per year. Assuming latrines are distributed
every 700 m and are 50 m2 in area (Swimley et al. 1998), N
deposition will range between 0.075 and 0.16 kg m-2 y-1. In
comparison, atmospheric wet N deposition in Alaska ranges between
0.00001 and 0.0002 kg m-2 y-1 (Lilleskov et al. 2001,
http://nadp.sws.uiuc.edu). Similarly, river otter latrines may
receive between 5 and 22 g m-2 y-1 of phosphorus. For comparison,
heath soils in northern Alaska contain 0.007–0.01 g m-2 P (Giblin
et al. 1991). Our recent studies suggest that otter fertilization
significantly alters plant community composition (Roe et al. In
prep) and increases photosynthetic capacity of the overstory layer
of coastal forests in the region (Roe et al. In review). Bowyer et
al. (1995) studied river otter habitat selection and home ranges in
the marine environment of PWS following the oiling of portions of
the sound from EVOS. Their habitat model showed that otters
strongly selected areas of old-growth forest in both the oiled and
nonoiled areas and preferred large rocks in oiled areas and
shallower tidal slopes in the nonoiled area. Otters in the nonoiled
area seemed to avoid commercially logged habitats, yet home ranges
were about twice the size for otters in oiled areas. Bowyer et al.
(1994) found significant declines in species richness and diversity
of otter food items in oiled versus nonoiled areas. These data
suggest that oil contamination can be detrimental to otters and
cause a reduction in their population (Bowyer et al. 2003). Such
reduction in otter numbers will likely reduce N and P fertilization
of coastal forests. Monitoring population status of river otters in
coastal environments may be crucial for evaluating the carbon
sequestration capacity of Alaskan coastal forests. In previous
studies, density estimates of river otters have relied on home
range calculations derived from radio telemetry data (Larsen 1983,
Melquist and Dronkert 1987). Following EVOS, Testa et al. (1994)
estimated otter densities in western PWS using a mark-recapture
technique with scats containing radioisotopes (from implants placed
in captured animals) supplemented with movement data from
radiomarked animals. Their estimates were 36 to 42 otters/100 km of
coastline for the oiled site at Herring Bay on Knight Island and 32
to 44 otters/100 km in the nonoiled site at Esther
6
-
Passage. However, the 95% confidence intervals for those
estimates overlapped. The inaccuracy of using radio telemetry data
alone and the restrictions of using radioisotopes in wild animals
make these techniques unsuitable for current estimates of otter
numbers. Monitoring the status of river otter populations is
challenging because their life history patterns make them difficult
to survey using traditional mark-recapture methods (Williams et al.
2002). Current methods for monitoring relative abundance involve
estimating the distribution and use of latrine sites (Bowyer et al.
2003) and the deposition rates of scats (i.e., scats
deposited/site/day). Using these measures it could be possible to
monitor population levels and trends among different areas of
coastline as long as these indirect measures are correlated with
formal population estimates. In addition, because the distribution
and use of latrines depends on available habitat, habitat use and
availability (as established by Bowyer et al. 1995) need to be
measured concurrently with monitoring fecal deposition rates.
Modern techniques for extracting and analyzing DNA from river otter
scats were recently developed (Hansen et al. 2008). These
procedures use microsatellite DNA extracted from freshly deposited
scats to identify individual otters based on DNA contained in cells
sloughed from their intestinal lining (Hansen et al. 2008).
Microsatellites are hypervariable, noncoding regions of short
repeats within DNA that vary in size. They can serve as genetic
markers because the regions may be amplified with specific
microsatellite primers and their sizes compared among individuals
with the aid of polymerase chain reaction products (Foran et al.
1997). Such methods provide identification of individual animals
(Blundell et al. 2002; Hansen et al. 2008). An increasing number of
studies have successfully employed fecal DNA analyses to obtain
formal population estimates for Eurasian otters (Lutra lutra;
Dallas et al. 2003, Janssens et al. 2007, Prigioni et al. 2006),
although few have been completed for North American river otters
(Guertin et al. In review). Nonetheless, because fecal DNA analyses
are expensive (estimated at $100 per sample after collection),
developing protocols that will provide reliable indices of
abundance is crucial for a relatively inexpensive monitoring
program.
This report summarizes the activities in calendar years 2004 –
2009 of a multi-agency collaborative project. Surveys in Kenai
Fjords National Park (KEFJ), Katmai National Park and Preserve
(KATM), Lake Clark National Park and Preserve (LACL), Prince
William Sound (PWS) and the Kodiak Island Archipelago (KOD) were
conducted under separate study plans and funding, but utilize the
same methodologies to meet similar goals. All projects and datasets
are presented here for comparative purposes and to better
illustrate the regional scale of this effort.
7
-
Methods Latrine site surveys We surveyed the KEFJ coastline for
latrine sites during July 5-10, 2004 and June 25-29, 2005 (Figure
1). In 2006, we also conducted a survey of random sites in that
location to assess habitat availability during June 20-24 (Figure
2). In KATM, a latrine site survey was conducted during July 2-10,
2005 (Figure 3), and in LACL during July 25-29, 2006 (Figure 4). In
PWS, we conducted a sound-wide survey August 9-21, 2004 (Figure 5)
and a more localized survey in Herring Bay, Lower Passage, and
Eleanor Island between May 25 and August 15, 2006 (Figure 6). We
surveyed latrines in KOD during 13–25 June, 2007 (Figure 7). The
length of shoreline surveyed at each location ranged from 60-945 km
and the number of occasions varied from 1 in KEFJ in 2004, KATM in
2005, and KOD in 2007, to 9 in PWS in 2006. We conducted a second
sampling of latrine sites to measure scat deposition rates and
collect fresh feces 48 hours after the first sampling occasion in
PWS in 2004, 24 hours afterward in KEFJ in 2005, and 5-7 days
afterward in PWS in 2006. In all but the LACL and PWS 2006 surveys,
a large vessel (M/V Serac in KEFJ and KATM, and the M/V Babkin in
PWS and KOD) served as a mobile camp. The surveys in LACL and in
PWS 2006 conducted from shore-based camps. Surveys were conducted
with 2-3 skiffs and a crew of 7-10 people. In all surveys, every
effort was made to locate and visit all potential latrine sites. We
recorded the location of each positive site (i.e., containing at
least 10 total scats or new scats) with a handheld GPS unit. In PWS
in 2004, in areas where surveys were conducted during previous
efforts (northern Knight Island: Herring Bay and Lower Passage),
Dangerous Passage (Jackpot, Ewan, and Paddy Bays and the western
coast of Chenega Island), Esther Passage, Port Gravina, and Orca
Bay (Olsen, Parshas, Sheep, Simpson, and Windy Bays), known
latrines were re-visited and otter activity was evaluated. Habitat
features of latrines and random sites At each new site in each
area, habitat features within a 10-m radius of the main entrance
from the water were evaluated and recorded following methods
described by Bowyer et al. (1995; 2003). These included aspect,
exposure to wave action, slope of the vegetated and tidal areas (in
degrees), composition of intertidal substrate ranked based on
percent cover of sand, gravel, small and large rock and bedrock,
composition of vegetation cover based on percent overstory and
understory vegetation and old growth trees, and potential burrow
sites. To estimate habitat availability in KEFJ, we sampled
approximately 660 km of the Park coastline from Bear Glacier to
Yalik Point excluding islands (Figure 2). To ensure that data
collection on random sites was equivalent to that on latrines, we
divided the coastline into 20-m segments or potential random sites.
This resulted in 32,993 sites along the 660-km coastline. We
estimated that we had sufficient funds and time to
8
-
sample approximately 400 of those sites. Because we lacked
covariates to describe anticipated habitat variability along the
coastline, we established a systematic array of evenly spaced
groups of potential random sites. Each group contained 82 sites
except the last group, which contained 29 sites. This resulted in
the establishment of 403 groups out of the 32,993 potential random
sites for an overall sampling effort of 1.2%. We used
stratified-random sampling to select a single random site from
within each group. To maintain consistent sampling probability, we
added 53 ghost sites to the last group of 29 sites and randomly
sampled from its new total of 82 sites. Although this created the
potential for missing real sites from this segment, we did sample a
site from this group. For each of these random sites, we measured
the same habitat features estimated at latrines within a 10-m
radius of the GPS location.
Figure 1. Distribution of river otter latrines along the coast
in KEFJ. Survey was conducted July 5-10, 2004 and encompassed 354
km of shoreline. Ninety four of these sites were re-visited in 2005
from June 25-29. Each site was visited twice representing one
occasion of “mark” and one of “recapture”.
9
-
Figure 2. Location of random sites sampled for habitat features
in KEFJ in June 2006.
Scat monitoring and collection At each site, the number of feces
was counted and all fresh feces, characterized by their distinct
glossy appearance and strong smell, were collected and preserved in
100% ethanol (EtOH). Each sample was collected in a separate a
50-ml sterile tube. All samples were labeled and stored in coolers
packed with glacial ice. In those surveys where we conducted 2 or
more sampling occasions, all scats found on the initial survey day
were marked with colored glitter and any new unlabeled scats were
counted in the following occasion. Relative abundance was measured
by the distribution and number of active latrine sites and fecal
deposition rate (Table 1). The latter was calculated based on the
number of unmarked scats found on each site divided by the number
of elapsed days between inspections (i.e., scats/ latrine site/
day).
DNA analyses Prior to DNA extraction, we sieved each fecal
sample through fine-mesh stainless steel, autoclavable sieves to
ensure the removal of all hard parts of prey material. This is an
important step for improving the quality and quantity of
extractable DNA through reducing the extraction and amplification
of non-target DNA. Sieving also helps avoid
10
-
the problem of the uneven distribution of cells shed through the
intestinal lining as documented by Kohn et al. (1995). Excess EtOH
was evaporated from each sample after sieving in a closed hood
(Hansen 2004). Following the sieving, we used a 200 μl sub-sample
to extract DNA with a QIAamp DNA Stool Mini Kit (Hansen et al.
2008).
Figure 3. Distribution of river otter latrines along the coast
in KATM. Survey was conducted July 2-10, 2005 and encompassed 168
km of shoreline.
11
-
Figure 4. Coastline surveyed for river otter latrine sites in
Lake Clark National Park in July 2006. Of the official coastline
measurement of 198 km for LACL, we surveyed approximately 50 km
(25%) within park boundaries between Difficult Creek in Tuxedni Bay
and Glacier Spit in Chinitna Bay. We also surveyed an additional 10
km along the southwest shore of Chisik Island, which is contained
within the Alaska Maritime National Wildlife Refuge.
12
-
Figure 5. Length of surveyed coastline in Prince William Sound
(in red), and distribution of otter latrines in the surveyed area
(yellow). Survey was conducted August 9-21, 2004 and encompassed
945 km of shoreline.
We performed DNA amplifications (PCR) using a PTC-0200 DNA
Engine Peltier Thermal Cycler (MJ Research, Inc., Waltham, MA).
Primers RIO-01, RIO-05, RIO-17, RIO-19, and RIO-20 developed for
river otters (Beheler et al. 2004; 2005), and LUT-701, LUT-733,
LUT-801 and LUT-829 developed for Eurasian otters (Lutra lutra;
Dallas and Piertney 1998) were used in PCR reactions. Positive
(blood samples from river otters with known genotypes) and negative
(PCR blank) controls were included with each PCR run to ensure the
reliability of PCRs and monitor contamination (Hansen et al. 2008).
Successful PCR reactions were resolved on an ABI 3130xl Automated
Sequencer (Applied Biosystems Foster City, CA; ABI) with
formamide-Liz ladder as an internal size standard for each lane.
Products were analyzed using ABI analysis software GeneMapper v4.
Successful amplification was determined by the presence of PCR
product of the expected size (Hansen et al. 2008). We obtained a
consensus genotype from positive PCRs with identifiable alleles
that had sufficient replication (Goossens et al. 2000). To reduce
genotyping error and time spent trying to amplify poor quality
samples, each sample that did not amplify after three PCR runs with
the three most reliable markers (RIO-19, LUT-733 and LUT829) was
discarded (Morin et al. 2001, Paetkau 2003). We evaluated genotypes
after two initial
13
-
reactions (Frantz et al. 2003, Hansen et al. 2008). Loci that
amplified the same heterozygous individual twice were recorded, and
homozygote genotypes were accepted on a provisional basis after a
stepwise amplification approach of up to seven PCRs. In the case
that an allele amplified only once to yield one heterozygote
genotype in seven runs with the other six runs resulting in the
same homozygous genotype, we designated the allele as constituting
a half-genotype (Frantz et al. 2003). Effects of habitat,
environmental conditions, diet, and parasite load on genotyping
success To optimize field collection of samples, we explored which
variables affected genotyping success of feces in KEFJ and PWS.
First, we used logistic regression with genotyping success as the
dependent variable (successful coded as 1 and unsuccessful coded as
0) and habitat features as the independent ones to determine if any
measures of shading or exposure are correlated with DNA degradation
in feces. Because we collected multiple samples at several of the
latrines, we called a site successful if at least one sample from
that site yielded usable DNA. We also evaluated whether
environmental conditions such as temperature and humidity were
correlated with amplification success. Nsubuga et al. (2004)
reported a negative relationship between the amount of DNA obtained
per sample and the ambient temperature at the time of fecal
collection for wild mountain gorillas (Gorilla beringei beringei).
Lucchini et al. (2002) found that amplification success of wolf
(Canis lupus) scats was higher for samples collected in the winter
than those collected in the summer. Farrell et al (2000) found that
carnivore feces collected in the rainy season had a much lower
amplification success than those collected in the dry season. Thus,
establishing a relation between genotyping success and
environmental conditions or habitat features will allow early
screening of samples and will result in significant savings in
effort and costs. We obtained data on average daily temperature and
relative humidity from the weather stations in Cordova, Seward,
Valdez, and Whittier. To explore the effects of environmental
conditions on genotyping success we used regression analyses with
either temperature, or humidity, or their multiplication as the
independent variable and percent of samples that were successfully
genotyped on each day as the dependent variable. Similarly, because
other studies noted the effects of diet on genotyping success of
feces (Hansen 2004; Murphy et al. 2002), we evaluated the effects
of diet on genotyping success by submitting prey remains sieved
from successful and unsuccessful fecal samples (n = 100) to Pacific
Identification Lab (University of Victoria, Victoria, British
Columbia). Samples that contained parasites and samples that were
free of parasites were included in this analysis because our
preliminary observations indicated lower genotyping success in
infected feces. We compared the effects of prey remains on
genotyping success with the binomial test (Zar 1999).
14
-
Observer bias Because it is likely that future monitoring
surveys will be conducted by different park personnel, we evaluated
whether genotyping success varied among samples collected by
different observers. Because the decision to collect a sample or
not is subjective based on the smell and appearance, differences
between observers are likely. We calculated the proportion of
successful samples for each observer and used a Z-test for multiple
proportions (Zar 1999) to identify observer bias.
Figure 6. Distribution of river otter latrine sites in Herring
Bay, Lower Passage, and Eleanor Island in Prince William Sound.
Survey was conducted May 25 – August 15, 2006 and encompassed 145
km of shoreline.
15
-
Figure 7. Distribution of river otter latrine sites in Big Bay
(Shuyak Island), Blue Fox Paramanof, Foul, and Malina Bays (Afognak
Island), Uganik Bay and Chiniak Bay/Ouzinkie Narrows (Kodiak
Island). Survey was conducted 13–25 June, 2007 and encompassed 376
km of shoreline.
Reliability of genotyping results We strictly adhered to the
comparative multiple tubes approach for assigning consensus
genotypes (Frantz et al. 2003, Hansen et al. 2008). We computed
genotyping error rates
16
-
(false alleles and allelic drop-out) based on the final sample
dataset with complete multilocus genotypes according to the
formulae in Broquet and Petit (2004) and Prugh et al. (2005). To
ensure that we used a sufficient number of loci for individual
identification, we calculated the probability of identity (PID; the
probability that two individuals drawn at random from the
population share identical genotypes at all typed loci) using
GIMLET 1.3.2 (Valière 2002). We calculated the theoretical upper
and lower limits of PID (Waits et al. 2001). The lower limit,
PID-unbiased, assumes a randomly mating population of unrelated
individuals in Hardy-Weinberg equilibrium (HWE). The upper limit,
PID-sib, assumes the population to be composed only of siblings,
and should be 0.01 or less if data are to be used for population
estimation (Mills et al. 2000). Population genetics We assessed
assumptions of random mating and tested for departures from HWE
using the probability test by Guo and Thompson (1992) as
implemented in GENEPOP 3.4 (Raymond and Rousset 1995). We used
FSTAT 2.9.3.2 (Goudet 2000) to calculate Weir and Cockerham (1984)
F-statistics and to test for linkage disequilibrium. Cavalli-Sforza
chord distances were calculated for each pair of populations
(Cavalli-Sforza and Edwards 1967) using the module GENDIST in
PHLYIP 3.67 (Felsenstein 1989). These distances were then used to
construct a tree diagram using Saitou and Nei’s (1987) neighbor
joining method with 1000 bootstrap replications implemented in the
modules NEIGHBOR and CONSENSUS in PHYLIP. Population
differentiation in each of the sampling areas was further assessed
using a Bayesian model implemented in STRUCTURE 2.2 (Pritchard et
al. 2000). We assumed an admixture model with correlated allele
frequencies (Falush et al. 2003). To estimate the number of
subpopulations (K), we performed 20 independent runs of K = 1-6
with a burn-in period of 100,000 followed by 100,000 Markov chain
Monte Carlo (MCMC) repetitions. We determined the most probable
number of subpopulations based on the mean log-likelihood of K
(L(K)), as well as ΔK, a measure of the second order rate of change
in likelihood of K (Evanno et al. 2005). We performed a final run
at the inferred K (100,000 burn-in and 500,000 MCMC repetitions)
and assigned individuals to a subpopulation based upon their
highest proportion of membership (q). We chose a threshold value of
0.70 to assign individuals to subpopulations (i.e., ≥ 70% of
ancestry can be attributed to the respective subpopulation;
Pritchard et al. 2000). Finally, to assess the degree of
immigration and emigration among the sampled populations, we used
the program MIGRATE 2.1.3 (Beerli and Felenstein 2001). We used the
Brownian motion model following parameters suggested by the
authors. We increased parameter stringency on consecutive analyses
to include longer runs and more recorded trees. Using information
on genetic differentiation, we delineated sub-populations and
calculated latrine densities, fecal deposition rate
(scats/site/day) and minimum number known alive (MNKA) for each
sub-population.
17
-
Estimation of abundance In this analysis only data from PWS 2006
were used because recapture rates in all other datasets were
insufficient. Because male river otters are wide ranging
(especially during the mating season; Blundell et al. 2002), a
significant portion of our dataset may consist of transient males.
Furthermore, because we sampled portions of the shoreline of 3
islands in the northern Knight Island complex (Figure 6), we likely
introduced some degree of capture heterogeneity by detecting otters
with home-ranges overlapping the edges of the study area. Finally,
because the duration of our survey included natal dispersal of
young otters from their dens we likely sampled new individuals
recruited into the population (Crait et al. 2006). Consequently,
the conditions of our study likely violate assumptions of
geographic and demographic closure required for analysis with
closed population CMR models (Amstrup et al. 2005). Therefore, we
used the Cormack-Jolly-Seber (CJS) open population model in program
MARK to obtain estimates of apparent survival (φ) and recapture
probability (p; Lebreton et al. 1992, Pledger et al. 2003, Pollock
2000, Seber 1982, White and Burnham 1999). The assumptions of this
model require that (1) animals have equal recapture probability,
and (2) apparent survival probability is homogenous between all
animals (Pledger et al. 2003, Pollock 2000, Seber 1982). Because we
lacked information on age and sex for otters in our dataset we
could not explore models that incorporated these covariates.
Instead, we compared models in which apparent survival (φ) and
recapture probability (p) either varied through time or were
constant. We evaluated model fit based on the Akaike Information
Criterion (AIC) adjusted for small samples, which indicates the
most likely model based on parsimony and optimal precision while
attempting to minimize bias (Burnham and Anderson 2002). We then
extrapolated river otter abundance in each occasion based on the
formula:
= (1)
where is abundance and nj is the number of animals marked (or
captured) at the jth occasion, while represents capture probability
at time j of animals marked in the previous occasion (Amstrup et
al. 2005). In order to assess lack of geographic and demographic
closure on population estimates, we first conducted the analyses
using all individuals identified from their fecal genotypes. We
then repeated the analyses but omitted any individuals that were
encountered only once. We incorporated samples that were
encountered at least twice (including individuals captured within
the same occasion – i.e., an animal encountered once on June 7 on
latrine HB06070, and again on June 8 at nearby LP06005), because
individuals observed only once likely represented transients
(introducing heterogeneity in p; Pradel et al. 1997). For both
datasets, we also calculated capture (p) and recapture (c)
probabilities and abundance (N) with closed-population models in
program MARK. We used AIC model selection procedures to select
among competing models as described by Burnham and
18
-
Anderson (2002). We then estimated river otter density by
accounting for the length of the shoreline surveyed. Assessing the
effect of sampling occasions on abundance estimates To assess the
effects of sampling occasions on the bias and precision of
abundance estimates we calculated estimates of apparent survival
(φ) and recapture probability with Cormack-Jolly-Seber (CJS) open
population model for the full and residents only datasets from PWS
2006. In addition, we calculated capture (p) and recapture (c)
probabilities and abundance (N) with closed-population models for
these datasets. We repeated these analyses and sequentially
truncated occasions from both datasets. For example, we obtained φ
and p for 8 occasions and then for 7 occasions, until only data
collected in the first 4 occasions were introduced into the models.
We then plotted the population estimates with 95% confidence
intervals to determine the number of occasions required to obtain
the least biased and most accurate abundance estimate. Relating
latrine density and fecal deposition rate to minimum number known
alive
To quantify the relation between otter abundance and measures of
relative abundance, we used multiple regression procedures with
latrine density and fecal deposition rate as the independent
variables and MNKA as the dependent one (Zar 1999). We employed
this measure of abundance rather than formal population estimates
because we were unable to generate the latter for most of our
sampling areas. For this analysis we used data from all
sub-populations in KEFJ, PWS, and KOD as well as data gathered for
KATM. In this analysis we also included data collected in a
companion study on Vancouver Island, British Columbia (Guertin et
al. In review). Results Latrine site surveys In KEFJ, we identified
162 latrine sites along the sections of coastline we surveyed in
2004. Of these, 153 were active (i.e., contained new scats or at
least 10 scats; Figure 1). This number of active latrines
represents an average density of 0.43 latrines per km of shoreline
(Table 1). Latrine densities appeared higher at the mouths of bays
away from the glaciated heads of the bays (Figure 1). Among the
active sites we counted 5,046 old and 297 new (24 hrs old) scats.
The maximum number of old scats found on a site was 235 and the
maximum number of new scats on a site was 20. Based on these
counts, we estimated that fecal deposition rate in KEFJ in 2004 was
32.98 feces per site for old scats and 1.94 fresh scats per site.
We collected 267 fresh scats for DNA analysis. Of the 153 latrine
sites sampled in KEFJ in 2004, 94 were active in 2005. A total of
4,912 old feces were counted and marked, and 416 new feces were
collected (total of 260 in first and 156 on second visit for a
deposition rate of 2.8 and 1.8 scats per site per day,
respectively; Table 1).
19
-
In 2006, we were able to complete our sampling of the habitat
characteristics for all pre-selected 403 random sites in 5 days
(Figure 2). During the survey of the KATM coastline, 58 river otter
latrines were identified (latrine density of 0.35 sites per km) and
a total of 63 fresh fecal samples were collected (deposition rate
of 1.1 scats per site per day; Table 1). Our inventory of the
shoreline of LACL in 2006 resulted in the location of only 4
latrine sites that collectively contained 11 old scats. These scats
were unsuitable for DNA analysis. Much of the coastline consisted
of muddy tidal flats that were not compatible with habitat
characteristics known to be favored by river otters. In PWS, we
sampled 286 latrine sites in 2004 (Figure 5). Of those sites, 109
were new sites (i.e., unknown from previous surveys). On the
recapture occasion we re-sampled 254 of these sites that we
considered active. This number of active latrines represents an
average density of 0.27 latrines per km of shoreline. The lowest
density of latrines was observed in Valdez Arm and Culross Pass
(Figure 5). Among all latrine sites we counted 1,048 new scats and
saved 302 of those for DNA analyses. For the recaptures, we counted
458 new scats and saved 263. Based on these counts, fecal
deposition rate of new scats in PWS is 4.13 feces per site for the
capture session and 1.80 fresh feces per site for the re-capture
session (Table 1). A total of 565 fresh scats were collected for
DNA analysis. In 2006 we identified 320 river otter latrines along
the shoreline of Herring Bay (HB), Lower Passage (LP), and Eleanor
Island (EI). We used stratified random sampling with bay (HB, LP
and EI) and number of feces (
-
genetic profiles at seven or more loci for 113 samples,
representing 106 individuals, which translates to a success rate of
20.0%. From 695 total fecal samples collected in KEFJ in 2004 and
2005 we obtained consensus genotypes at seven or more loci for 120
samples representing 103 individuals and yielding a 17.3% success
rate. For PWS 2006 we generated 261 genetic fingerprints of 131
individuals from an original sample set of 963 for a success rate
of 27.1%. In KATM, 15 of 63 samples were successfully genotyped (or
23.8%) representing 12 unique individuals, and in KOD, 82 unique
individuals were represented in the sample of 94 samples that
yielded consensus genotypes. In KOD success rate was the highest at
36% (94 of 261 samples). The probability of obtaining an incorrect
multilocus genotype after replication at all eight microsatellite
loci ranged from 0.0004 to 0.009 across all populations. Allelic
richness and heterozygosity were high in all areas except KOD
(Table 2). The probability that two individuals drawn at random
from a given population share identical genotypes at all loci was
low (KEFJ – 1 in 15,948,963; PWS – 1 in 7,575,758; KATM – 1 in
333,333; KOD – 1 in 38,760). Effects of habitat, environmental
conditions, diet, and parasite load on genotyping success Despite
large variation in site characteristics, no habitat feature could
explain the differences we observed in genotyping success of river
otter fecal samples (Figures 8 – 11; Logistic regression, P >
0.316). We found no relation between overstory or understory cover
and genotyping success despite the potential of such cover to
reduce exposure of samples to direct sunlight. Similarly, we found
no relation with slope, occurrence of bedrock, or aspect.
overstory
120100806040200-20
SU
CC
ES
S
1.2
1.0
.8
.6
.4
.2
0.0
-.2
Figure 8 – Distribution of percent overstory in relation to
successful genotyping (coded 1) and unsuccessful genotyping (coded
0) for river otter fecal samples collected in KEFJ in July
2004.
In contrast, genotyping success declined with an increase in
temperature (Figure 12; R2 = 0.22, P = 0.05). Genotyping success
was higher on days with temperatures bellow 16˚C (44.2% ± 2.5),
compared with days when the temperature exceeded 16˚C (30.2% ± 3.7;
ANOVA, P = 0.005). This suggests that samples collected on hot days
should be discarded to reduce efforts and costs of DNA
analyses.
21
-
understory
120100806040200-20
SUC
CES
S1.2
1.0
.8
.6
.4
.2
0.0
-.2
Figure 9 – Distribution of percent understory cover in relation
to successful genotyping (coded 1) and unsuccessful genotyping
(coded 0) for river otter fecal samples collected in KEFJ in July
2004.
VEGSLOP
706050403020100
SU
CC
ES
S1.2
1.0
.8
.6
.4
.2
0.0
-.2
22
Figure 10 – Distribution of vegetative slope in relation to
successful genotyping (coded 1) and unsuccessful genotyping (coded
0) for river otter fecal samples collected in KEFJ in July
2004.
ASPECT
1086420
SU
CC
ES
S
1.2
1.0
.8
.6
.4
.2
0.0
-.2
Figure 11 – Distribution of aspect as measured in 8 compass
directions in relation to successful genotyping (coded 1) and
unsuccessful genotyping (coded 0) for river otter fecal samples
collected in KEFJ in July 2004.
-
Table 1. Shoreline lengths and number of river otter latrine
sites surveyed among different bays in southwestern and
southcentral Alaska between 2004 - 2007. Latrine densities were
calculated based on the number of sites located along the coast.
Fecal deposition was estimated based on the number of scats counted
per site in each bay.
Area Bay Year Shoreline
length Number of
latrines Latrine density
Scats/site/day Minimum number known alive
Katmai 2005 168 58 0.35 1.10 12
Kenai Fjords Aialik Cape 2004 213 106 0.50 2.17 39
Nuka Bay* 2004 129 56 0.43 1.31 4
Aialik Cape 2005 213 106 0.50 2.76 39
Nuka Bay 2005 129 56 0.43 3.72 21
Kodiak Big 2007 22 23 1.05 3.17 15
Blue Fox 2007 27 24 0.89 1.21 15
Paramanof/Foul/ Malina 2007 158 71 0.45 1.00 29
Uganik 2007 92 49 0.53 1.35 21 Prince William Sound Jackpot 2004
154 159 1.03 1.88 15
Olson 2004 184 93 0.51 0.83 13 Valdez Entrance 2004 154 43 0.28
3.37 17 Unakwik Inlet 2004 207 146 0.71 4.77 28
Herring@ 2006 76 138 1.81 2.69 20
Lower Passage@ 2006 46 88 1.90 0.97 7
Eleanor Island@ 2006 21 36 1.72 0.82 9 *Data excluded from
further analyses because of low genotyping success
@Only data from the first occasion were included to ensure
compatibility with the other datasets
23
-
Diet analyses revealed that genotyping success was lower for
samples containing the remains of Cottidae and Stichaeidae
(Binomial test, P < 0.05; Table 2). Samples containing the
remains of Syngnathidae yielded no consensus genotypes. No other
fish remains appeared to significantly affect genotyping success
although samples containing Pleuronectiformes, Salmonidae and
Scorpaenidae exhibited consistently higher genotyping success.
Thus, evaluation of diet composition may assist in pre-screening of
fecal samples for DNA analyses. Through sieving we found a
difference in the incidence of intestinal parasites in river otter
feces from KEFJ (parasites occurred in 36% of samples) and PWS
(parasites occurred in 10% of samples). The occurrence of parasites
negatively affected genotyping success. Whereas 49.1% of samples
that were free of parasites yielded usable DNA, only 19.8% of
infected ones were successful (Z-test of 2 proportions, P <
0.01). Observer bias We detected no observer bias (Z-test of
multiple proportions, P = 0.38). On average 38% (± 8%) of samples
collected by HH yielded consensus genotypes. Similarly, 42% (± 9%)
collected by MBD, and 49% (± 9%) collected by KEO yielded consensus
genotypes. Thus, it will be possible to assign the task of sample
collection to multiple observers in future monitoring efforts.
10 12 14 16 18 20 22 2410
20
30
40
50
60
Temperature (°C)
Gen
otyp
ing
succ
ess
(%)
Figure 12 – Genotyping success (percent of samples that yielded
DNA from the total collected on each day) declined with increasing
temperatures. Data was collected in KEFJ and PWS 2004.
Population genetics In the following analyses we included data
obtained by Guertin et al. (In review) for river otters inhabiting
the southern part of Vancouver Island, British Columbia (BC),
Canada (Table 3). Pairwise FST values suggested significant
differentiation among all sampled populations (Table 4).
Cavalli-Sforza chord distance with BC rooted as an out-group
indicated strong divergence of KEFJ and PWS from KOD and KATM, as
well as well-supported divergence between KEFJ and PWS. In
contrast, divergence of KOD
24
-
from KATM was weak, with less than half of the projected trees
supporting separation of the 2 populations (Figure 13). These
results suggest that KATM and KOD otters likely differentiated from
one ancestral stock that inhabited the Pleistocene southwestern
shores of Alaska, and was isolated from other more easterly
populations by distance. Table 2. Percent occurrence of prey items
in scats that yielded consensus genotypes (Successful) and in scats
that failed to produce consensus genotypes (Unsuccessful).
Statistical comparisons were conducted only for families that
occurred in at least 10 of the 104 analyzed samples (*). Bolded
numbers represent overall values for families with multiple
species.
Family Common name Scientific name Percent occurrence Successful
UnsuccessfulAgonidae Poacher Agonidae 0.00 1.18 Ammodytidae* Sand
lance Ammodytes hexapterus 10.53 9.41 Anarhichadidae Wolf eel
Anarrichthys ocellatus 0.00 1.18 Bathymasteridae Searcher
Bathymaster signatus 5.26 0.00 Searcher Bathymaster sp. 0.00 1.18
Clupeidae Pacific herring Clupea pallasi 0.00 1.18 Cottidae* Padded
sculpin Artedius fenestralis 5.26 9.41 Scalyhead sculpin Artedius
harringtoni 5.26 5.88 Smoothhd sculpin Artedius lateralis 10.53
3.53 Buffalo sculpin Enophrys bison 0.00 2.35 Leister sculpin
Enophrys lucasi 0.00 1.18 Gymnocanthus Gymnocanthus sp. 5.26
0.00
Red irish lord Hemilepidotus hemilepidotus 0.00 10.59 Irish lord
sp Hemilepidotus sp. 15.79 15.29 Shorthorn sculpin Myoxocephalus
scorpius 0.00 1.18 Great-type sculpin Myoxocephalus sp. 0.00 4.71
Tidepool sculpin Oligocottus maculosus 0.00 2.35 Sculpin
Oligocottus sp. 0.00 9.41 Sculpin Cottidae 0.00 5.88 42.11a 71.76
Gadidae* Saffron cod Eleginus gracilis 0.00 1.18 Pacific cod Gadus
macrocephalus 0.00 2.35 Tomcod Microgadus proximus 5.26 1.18
Pollock Theragra chalcogramma 0.00 1.18 Gadid (not hake) Gadidae
0.00 3.53 5.26 9.41 Gasterosteidae Three-spined stickleback
Gasterosteus aculaeatus 0.00 1.18 Hemitripteridae Crested sculpin
Blepsias bilobus 0.00 1.18 Silverspot sculpin Blepsias cirrhosus
0.00 1.18 Sailfin sculpin Nautichthys sp. 0.00 1.18
Hexagrammidae* Kelp greenling Hexagrammos decagrammus 10.53
11.76
Rock greenling Hexagrammos lagocephalus 0.00 3.53 W-s greenling
Hexagrammos stelleri 0.00 1.18 Greenling Hexagrammmos sp. 0.00 3.53
10.53 20.00 Liparididae Snailfish Liparidinae 0.00 3.53 Pholidae*
Crescent gunnel Pholis laeta 5.26 30.59
25
-
Red gunnel Pholis schultzi 0.00 1.18 Gunnel Pholididae 15.79
7.06 21.05 38.82 Pleuronectiformes* Arrowtooth flounder Atheresthes
stomias 5.26 1.18 Pacific sanddab Citharichthys sordidus 0.00 1.18
Sanddab Citharichthys sp. 0.00 1.18 Halibut Hippoglossus stenolepis
5.26 0.00 Rock sole Lepidopsetta sp. 21.05 20.00 Dover sole
Microstomus pacificus 0.00 2.35 English sole Parophrys vetulus 5.26
0.00 Starry flounder Platichthys stellatus 0.00 1.18 C-o turbot
Pleuronichthys coenosus 0.00 1.18 Sand sole Psettichthys
melanostictus 5.26 1.18 Flatfish Pleuronectiformes 5.26 3.53 47.37
32.94 Salmonidae* Salmon Oncorhynchus sp. 47.37 32.94 Scorpaenidae*
Rockfish Sebastes sp. 26.32 17.65
Stichaeidae* Slender cockscomb Anoplarchus insignis 0.00 2.35
Cockscomb Anoplarchus sp. 0.00 11.76 Daubed shanny Lumpenus
maculatus 0.00 1.18 Snake prickleback Lumpenus sagitta 0.00 4.71
Arctic shanny Stichaeus punctatus 0.00 5.88 Black prickleback
Xiphister atropurpures 0.00 7.06 Rock prickleback Xiphister mucosus
5.26 0.00 Prickleback sp. Stichaeidae 0.00 3.53 5.26a 36.47
Syngnathidae* Bay pipefish Syngnathus leptorhynchus 0.00a 11.76
Zaproridae Prowfish Zaprora silensus 0.00 1.18 Cephalopoda Squid
unidentified 0.00 1.18
Polycaeta Polycaete unidentified 5.26 0.00 a significantly
different at p = 0.05
These results seemingly contradict those provided by the
STRUCTURE analysis. Log-likelihood estimates suggested the
existence of 4 distinct populations: KOD, KEFJ, PWS (with KATM
individuals included in that cluster), and BC. Values of ΔK,
however, implied the existence of only 3 populations, combining
KEFJ, KATM and PWS into a single cluster (Figure 14). Nonetheless,
none of the KEFJ individuals was mis-assigned to the PWS-KATM
cluster with probability greater than 0.5 and only 16 individuals
from PWS (or 6.7%) were mis-assigned to the KEFJ cluster with
probability greater than 0.7. The conclusion that river otters from
KEFJ are genetically distinct from those inhabiting PWS is further
supported by results from MIGRATE (Table 5). Migration rates across
all 5 populations were low and similar, suggesting little genetic
exchange among the study populations.
Similarly, FST values indicated that population sub-structuring
occurred within all sampled areas (Tables 6 and 7), although
STRUCTURE analysis detected sub-structuring only in KEFJ (Figure
15) with isolation between Resurrection Bay (in the Northeast)
to
26
-
Nuka Bay (in the Southwest; Figure 15). This was supported by an
FST value of 0.091 between Aialik Cape (AC) and Nuka Bay (NB; 95%
Confidence interval 0.036 - 0.159). Based on these observations, we
calculated all population parameters separately for each of the
sub-populations (Table 1). Table 3. Sample size (n) mean number of
alleles frequency (A ± SD), expected heterozygosity (He), observed
heterozygosity (Ho), and inbreeding coefficient (FIS) for five
river otter populations along the Pacific coast. Genotypes were
obtained from 471 fecal samples representing unique individuals.
Samples were collected between 2004 and 2007.
Location n A (SD) He Ho FIS
Kodiak 82 4.00 (1.86) 0.420 0.433 0.098b
Katmai 12 4.50 (1.69) 0.622 0.718a -0.168b
Kenai 103 6.13 (2.23) 0.683 0.674 0.013
Prince William Sound 237 6.63 (2.36) 0.659 0.695 -0.054b
British Columbia* 49 4.38 (2.00) 0.570 0.520a 0.106b
*Adopted from Guertin et al. (In review) a significantly
different than expected b significantly different than zero
Table 4. Pairwise FST values (upper diagonal) and 95% confidence
intervals (lower diagonal) for 5 river otter populations along the
Pacific coast (Figure1). Genotypes were obtained from 471 fecal
samples representing unique individuals. Samples were collected
between 2004 and 2007. Adopted from Seymour et al. (In prep).
KOD KATM KEFJ PWS BC KOD - 0.254 0.248 0.190 0.394 KATM
0.088-0.449 - 0.070 0.059 0.150 KEFJ 0.126-0.371 0.011-0.130 -
0.076 0.197 PWS 0.081-0.287 0.020-0.110 0.035-0.126 - 0.159 BC
0.160-0.472 0.083-0.153 0.100-0.314 0.047-0.279 -
Table 5. Migration rates (as calculated by program MIGRATE)
among 5 river otter populations along the Pacific coast (Figure1).
Genotypes were obtained from 471 fecal samples representing unique
individuals. Samples were collected between 2004 and 2007. Adopted
from Seymour et al. (In prep). Receiving population Source
Population KOD KATM KEFJ PWS BC KOD 1.155 0.957 1.572 0.725 KATM
1.209 0.761 1.814 0.986 KEFJ 1.306 1.260 1.900 1.378 PWS 1.878
1.656 1.852 1.138 BC 1.201 1.184 0.986 1.726
27
-
Figure 13 – Cavalli-Sforza chord distance tree (as calculated by
program PHILYP) with southern Vancouver Island, British Columbia
(BC) rooted as an out-group, indicated strong divergence of Kenai
Fjords National Park (KEFJ) and Prince William Sound (PWS) from the
Kodiak Island Archipelago (KOD) and Katmai National Park and
Preserve (KATM), high support for divergence between KEFJ and PWS,
and weak support for divergence of KOD from KATM. Genotypes were
obtained from 471 fecal samples representing unique individuals.
Samples were collected between 2004 and 2007. Adopted from Seymour
et al. (In prep).
28
-
Figure 14. Posterior probability assignments of river otters to
two genetic clusters inferred by STRUCTURE for 5 river otter
populations sampled along the Pacific coast between 2004 and 2007.
No individuals from the Kodiak Island Archipelago (KOD) or British
Columbia (BC) were mis-assigned to any other population. Animals
from Katmai National Park and Preserve (KATM) largely clustered
with those from Prince William Sound (PWS). Adopted from Seymour et
al. (in prep).
29
-
Table 6. Pair-wise genetic distances (FST upper diagonal, 95%
confidence interval lower diagonal) between subpopulations of river
otters in Prince William Sound (sound-wide survey in 2004 [a] and
Northern Knight and Eleanor islands in 2006 [b]). We designated
these areas based on distance and previous demarcation of
populations by Blundell et al. (2002). Northern Knight Island
complex (NKIC) encompassed Drier Bay, Herring Bay, Lower Passage,
Eleanor Island and Naked Island; Jackpot Bay (JB) included samples
collected in Lower Chenega Island, Icy bay, Jackpot Bay and Culross
Passage; Olson Bay (OB) included the area from Port Gravina to
Cordova; Valdez Entrance (VE) the area from Sawmill Bay east to
Port Fidalgo; and Unakwik Inlet (UI) from Ester Passage East to the
Columbia Glacier. Adopted from Ott et al. (In prep a). (a)
Study area NKICa Jackpot Bay Olson Bay Valdez Entrance Unakwik
Inlet
NKICa 0.062 0.086 0.046 0.035
Jackpot Baya 0.018-0.112 0.125 0.074 0.051
Olson Bay 0.039-0.134 0.066-0.181 0.024 0.026
Valdez Entrance 0.007-0.091 0.019-0.123 0.002-0.049 0.017
Unakwik Inlet 0.006-0.064 0.019-0.092 0.004-0.045
0.011-0.021
(b)
Study area Herring Bay Lower Passage Eleanor Island
Herring Baya 0.012 0.021
Lower Passagea 0.002-0.023 0.025
Eleanor Islanda 0.007-0.037 0.007-0.046 a Includes Herring Bay,
Lower Passage, Eleanor and Naked Islands
30
-
Table 7. Pairwise FST values (upper diagonal) and 95% confidence
intervals (lower diagonal) for river otter sub-populations along
the coast of the Kodiak Island Archipelago (Figure 7). Genotypes
were obtained from 80 fecal samples representing unique
individuals. Samples were collected in 2007. Adopted from Seymour
et al. (In prep).
Big Bay Blue Fox Bay Paramanof Bay Uganik Bay Big Bay 0.085
0.133 0.026 Blue Fox Bay 0.003-0.159 0.068 0.052 Paramanof Bay
0.053-0.226 0.008-0.132 0.054 Uganik Bay 0.002-0.126 0.023-0.094
0.027-0.091
Estimation of abundance These analyses were performed only on
the PWS 2006 dataset because there was an insufficient number of
true recaptures in the other datasets. In KEFJ and PWS 2004 and
2005, 17 individuals were observed twice (seven in PWS and ten in
KEFJ). Two individuals in KEFJ were observed three times and one
was observed four times. However, these recaptures usually occurred
on the same collection dates and in most cases on the same latrine
site (only six individuals were recaptured on the same day at
different latrine sites, and only two were observed on different
latrines on separate days). The 261 samples collected in PWS 2006
and for which we had genetic profiles at six or more loci
represented 131 individuals (minimum number known alive). Of these,
58 individuals were recaptured between two and eight times. Models
with constant apparent survival and recapture probability that
varied with time ([φ (.), p(t)]) had the best fit to our data in
both datasets (131 and 58) based on AICc and AIC weights,
regardless of whether we used the default or other link functions
(Table 8). The estimate of apparent survival for the full dataset
was 0.90, and 1.00 for recaptures only dataset (Table 9). Ben-David
et al. (2002) estimated survival rate of 1.0 for river otters in
PWS during a similar time period over the summer. Thus, we believe
we identified the resident population with our “residents only”
capture-recapture dataset. Recapture probabilities were generally
higher and similar in the dataset containing only animals that were
observed at least twice, although the temporal pattern of change in
recapture probabilities was similar in both datasets. In contrast,
the temporal pattern of the post hoc estimates of abundance
differed between the full dataset and the two datasets excluding
transients and edge-otters. Abundance estimates for the full
dataset ranged from 85 in occasion 6 to 182 in occasion 9 (a 114.1%
difference), while estimates from the recaptures only dataset for
the same occasions varied between 57 to 67 individuals (or a 17.5%
difference). Average density of otters in the areas is estimated at
1 otter per 1.18 km of shoreline for the full dataset and 1
resident otter per 2.33 km. The model with the best fit to the data
using closed-population models was one estimating one uniform
capture probability but two different recapture probabilities. This
was true when we estimated these parameters for the full dataset
(131 individuals; AIC weight 1.00) as well as the resident only
dataset (58 individuals; AIC weight 0.99; Table 10). As with open
population models, capture and recapture probabilities were
31
-
higher for the resident only dataset although the patterns were
similar for both. With these models abundance of river otters in
Herring Bay, Lower Passage, and Eleanor Island was estimated at 163
(144 – 207) individuals for the full dataset, and 58 (58 – 65) for
residents only. This translates to a density of 1 otter per 0.89 km
of shoreline. Excluding transients the density of otters in our
study area was 1 resident per 2.5 km of shoreline.
Ressure
ction B
ay
Aialik B
ay
Harris B
ay
Two Arm
Bay
McCarty
Fjord
Nuka B
ay
Num
ber o
f ind
ivid
uals
0
5
10
15
20
Population 1 Population 2
Figure 15. Number of individuals assigned to each sub-population
in KEFJ at a threshold of 0.7. Survey locations are arranged from
east to west with Resurrection Bay representing the Northeastern
most extent of our KEFJ study area and Nuka Bay in the Southwest.
Adopted from Ott et al. (In prep a).
32
-
Table 8. Open-population CJS models used to estimate river otter
abundance from capture-recapture data derived from fecal genotyping
of river otters in Herring Bay, Lower Passage, and Eleanor Island
in Prince William Sound, Alaska, USA. Samples were collected in
nine occasions between May 25 and August 15, 2006. Adopted from Ott
et al. (In prep b). . Model AICc Delta AICc AICc Weights Model
Likelihood Num. Par Deviance
[φ (.) p (t)]† 5322.5 0.0 1.000 1.000 9 1589.1
[φ (t) p (.)]† 5451.1 129.0 0.000 0.000 9 1718.1
[φ (.) p (.)]† 5547.0 224.5 0.000 0.000 2 1827.7
[φ (.) p (t)] ‡ 4174.1 0.0 0.975 1.000 9 1672.1
[φ (t) p (t)] ‡ 4181.5 7.3 0.025 0.025 15 1667.2
[φ (t) p (.)]‡ 4353.1 179.0 0.000 0.000 9 1851.1
[φ (.) p (.)]‡ 4462.6 288.5 0.000 0.000 2 1974.7 † = Full
dataset (131 individuals) ‡ = Recaptured otters only (58
individuals)
33
-
Table 9. Estimates of apparent survival (φ ± SE), recapture
probability (p ± SE) and mean population size (Ν ± 95% confidence
interval) produced by the best fit model ([φ (.), p(t)]), from
capture-recapture data derived from fecal genotyping of river
otters in Herring Bay, Lower Passage, and Eleanor Island in Prince
William Sound, Alaska, USA. Samples were collected in nine
occasions between May 25 and August 15, 2006. Adopted from Ott et
al. (In prep b).
Full dataset (131 individuals)
Recaptures only
(58 individuals)
estimate SE estimate SE
φ 0.904 0.012 1.000 1.39 x 10-6 p1-2 0.273 0.024 0.360 0.029
p2-3 0.199 0.018 0.270 0.022 p3-4 0.167 0.015 0.217 0.018 p4-5
0.376 0.023 0.490 0.022 p5-6 0.077 0.011 0.094 0.012 p6-7 0.173
0.017 0.203 0.017 p7-8 0.122 0.014 0.145 0.014 p8-9 0.159 0.018
0.179 0.015
N 123 99 – 147* 62 58 – 66*
*95% confidence intervals
Table 10. Estimates of capture (p ± SE), recapture probability
(c ± SE) and population size (Ν ± 95% confidence interval) produced
by the best fit model from capture-recapture data derived from
fecal genotyping of river otters in Herring Bay, Lower Passage, and
Eleanor Island in Prince William Sound, Alaska, USA. Samples were
collected in nine occasions between May 25 and August 15, 2006.
Adopted from Ott et al. (In prep b).
Full dataset (131 individuals)
Recaptures only
(58 individuals)
estimate SE estimate SE
p 0.163 0.030 0.392 0.046 C1 0.207 0.025 0.338 0.038 C2 0.082
0.013 0.155 0.025
N 163 144 – 207* 58 58 – 65*
*95% confidence intervals
34
-
Assessing the effect of sampling occasions on abundance
estimates For all datasets with truncated number of occasions the
best fit in open-population models was for constant apparent
survival and recapture probability that varied with time ([φ (.),
p(t)]). For closed population models the best fit was for one
uniform capture probability but two different recapture
probabilities, except for the 4 occasion dataset where the best fit
was a model with one capture and one recapture probabilities.
Population estimates for the full datasets with closed-population
models were higher than for the same dataset with open population
models only when the 9 occasions were included, suggesting that
data obtained in that occasion introduced bias into
closed-population estimates. This is likely a result of a
recruitment pulse that occurs at the end of summer, when young
otters join their dames foraging and scent marking at latrines.
There was little change in point estimates when the dataset was
truncated between 8 and 5 occasions using open-population models,
although confidence intervals were broader when only 5 occasions
were used to produce the estimates. The residents only dataset
exhibited similar patterns and here too precision was lower with
only 4 and 5 occasions. These analyses suggest that unbiased and
precise population estimates for coastal river otters can be
achieved with a minimum of 6 occasions (Figure 16). Relating
latrine density and fecal deposition rate to minimum number known
alive We found no relation between latrine density and fecal
deposition rate and MNKA (Multiple regression, R2 = 0.207, p =
0.156; p latrine density = 0.101, p fecal deposition = 0.523).
Similarly, we found no relation between fresh fecal deposition at
each occasion and the estimated number of otters for the PWS 2006
dataset (R2 = 0.001, p = 0.935). There was a marginally significant
relation between recapture probability and fecal deposition rate in
that dataset (R2 = 0.459, p =0.065), likely reflecting the fact
that genotyping success is a function of the number of samples
collected. Discussion Our results suggest that latrine density and
fecal deposition rates relative to minimum number known alive may
be poor predictors of the status of populations of coastal river
otters. Furthermore, because sampling of entire sections of
coastline is impractical, the best approach for monitoring river
otters should adhere to the requirements of open-population models.
Such models depend on a minimum of 3 sampling occasions. Our data,
however, indicate that 6 or more sampling occasions will be
required to obtain unbiased and precise estimates of river otter
abundance. Although the costs of monitoring otter abundance and
population status through time with fecal DNA analyses may be more
costly than relative abundance indices and may seem prohibitive,
this technique offers important benefits:
1. With fecal DNA analyses it is possible to delineate
meaningful study populations, because sub-populations and geneflow
between them can be identified. For example, using this extensive
dataset we were able to identify not only population level
differentiation between KOD, KATM, KEFJ and PWS but also identify
sub-
35
-
population structuring within KOD, KEFJ, and PWS. From our
analyses it appears that ocean currents may be responsible for much
of the isolation of river otter populations along the Alaska coast.
Strong ocean currents and tidal fluxes associated with the Alaska
Coastal Current (ACC) and Shelikof Straights likely limit dispersal
of otters between KOD and KATM (Seymour et al. in review). Strong
currents of the ACC in combination with a high frequency of storms
along the Kenai Peninsula potentially contributed to the patterns
we observed, on a smaller scale, within KEFJ. For example, Aialik
Cape may represent a geographic barrier to otter movement because
it is precipitously rocky and exposed to storms and swells from the
Pacific Ocean. In contrast, PWS is more sheltered from the ACC by
Hinchinbrook, Latouche, and Montague Islands (Bang and Moores
2003). Therefore, river otter populations within PWS may not be
subject to the extreme effects of the ACC and may experience
enhanced geneflow as suggested by the lower values of FST among
sub-populations (Ott et al. In prep a). It is not too surprising
that geneflow in semi-aquatic mammals such as river otters could be
influenced by ocean currents given that ocean currents can dictate
geneflow even in marine fishes (such as the rosethorn rock fish,
Sebastes helvomaculatus; Rocha-Olivares and Vetter 1999). Indeed,
our results parallel those reported by Almeida et al. (2005) for
the semi-aquatic Neotropical water rat inhabiting off-shore islands
in Brazil.
2. With fecal DNA analyses it is possible to identify important
source populations worthy of conservation. For example, our data
suggest that river otters produced in Paramanof, Foul, and Malina
Bay on Afognak Island may immigrate to Blue Fox Bay and replenish
that population, which is exposed to relatively high levels of
trapping (Golden et al. 2009).
3. By assessing geneflow between source populations it is
possible to identify geographical locations that serve as corridors
and thus merit special protection. For example, much of the
coastline in LACL consists of muddy tidal flats that are selected
against by river otters. In our survey of the shoreline we found
few latrines and none of the samples we collected yielded viable
DNA. Nonetheless, our results suggest that this area may be an
important corridor for geneflow among otter populations in KATM and
PWS. Thus, this area merits enhanced status of protection.
4. While monitoring river otter populations with fecal DNA
analyses it is possible to evaluate their effects on the carbon
sequestration capacity of Alaskan coastal forests. Recently we were
able to demonstrate that conifers growing on river otter latrines
have 2.45 times higher photosynthetic capacity than their
conspecifics growing on adjacent non-fertilized sites (Roe et al.
In review). By estimating fecal deposition rate at latrines we were
able to unequivocally link this higher photosynthetic capacity to
marine-derived nutrients brought to latrines by otters (Roe et al.
In review).
To reduce time, effort, and cost associated with monitoring
river otters with fecal DNA analyses, we recommend observers
preferentially collect samples that contain anal gland secretions.
Also, any sample should be discarded if it does not amplify after
three PCRs with the most reliable primers (e.g., RIO-19, LUT-733
and LUT829), and one or more of the following conditions apply: 1)
it contains parasites, 2) it contains remains of
36
-
37
Cottidae and Stichaeidae, or 3) it was collected when ambient
temperatures exceeded 16˚C. Our analyses demonstrated that there
was no observer bias in the identification of samples that would
yield DNA. Also, none of the habitat features that we measured
could explain genotyping success or failure. Thus, it will be
possible to train novice observers to identify otter latrines based
on habitat features as well as to collect high quality samples. It
is unfortunate that we were unable to formally estimate abundance
of river otters in all sampled populations. Our conclusion that
latrine densities and fecal deposition rate were poor predictors of
otter abundance relied only on values of MNKA, which likely do not
represent abundance (Amstrup et al. 2005). Unfortunately, to obtain
formal estimates, it is likely that we would have needed to conduct
at least 6 sampling occasions in each area at each survey. In PWS
2006, the first 6 occasions were conducted between May 25 and July
25, a period too long for most monitoring studies. Recently (June
2009), we sampled 60 river otter latrines in Herring Bay, Lower
Passage, and Eleanor Island over 11 consecutive days to explore
whether formal estimates can be obtained over a shorter time
period. Results from the 787 samples we collected in that effort
are pending. Despite our failure to obtain formal population
estimates for KEFJ, KATM, KOD, and most of PWS, our results provide
the first large-scale description of the distribution of river
otters in southcentral and southwestern Alaska. It is a first
glimpse at the range of naïve densities and the differences in such
measures between several geographic areas. It is also the first
rigorous attempt to assess the validity of indices of relative
abundance. Future efforts should be dedicated to refining of
monitoring protocols and their implementation in advance of
environmental change. Other accomplishments In 2005, we tested the
efficacy of deploying non-invasive hair snares in KEFJ. Between 1-3
snares were set on 48 latrines on the mark occasion and collected
24 hours later during the recapture occasion. Hair snaring success
was 1 capture per 3.6 trap-nights and yielded a total of 20 samples
for DNA analyses (DePue and Ben-David 2007).
-
Figure 16. Estimated number (± 95% confidence interval) from
capture-recapture data derived from fecal genotyping of river
otters in Herring Bay, Lower Passage, and Eleanor Island in Prince
William Sound, Alaska, USA. Samples were collected in nine
occasions between May 25 and August 15, 2006. (A) Estimates derived
from open-population models with the full dataset (131
individuals); (B) estimates derived from open-population models
with the residents only dataset (58 individuals); (C) estimates
derived from closed-population models with the full dataset (131
individuals); (D) estimates derived from closed-population models
with the residents only dataset (58 individuals).
38
-
Acknowledgements The following people were instrumental in data
collection and analyses: Shannon Albeke (UGA), Alan Bennett
(NPS-SWAN), Jessica Boyd (UW), Aaron Christ (ADFG), David Crowley
(ADFG), John DePue (UW), Greg Fahl (KATM), Matt Gray (KEFJ), Dan
Guertin (SFU), Heidi Hansen (UW), Michael Harrington (ADFG),
Matthew Hayes (UW), Jason Herreman (UW), Tahzay Jones (KATM), Mandy
Kauffman (UW), Jennie Lewis (UW), Dan Logan (USFS), Ian Martin
(KEFJ), Wayne Melquist (Idaho), Nathan Nibbelink (UGA), Kaithryn
Ott (UW), Aaron Poe (USFS), Todd Rinaldi (ADFG), Mathew Seymour
(UW), Carol Scott (Northstar Borough School District), Elizabeth
Solomon (ADFG), Michael Tetreau (KEFJ), Marci Trana (UW), and James
Wendland (ADFG). We thank the crew of the M/V Serac for providing
logistical support for our work in KEFJ and KATM. Similarly we
thank the crew of the M/V Babkin for their help in the PWS and KOD
surveys. Without the enthusiasm, willingness to help and
accommodate, and the wonderful food they cooked, this project could
not have been done. We very much appreciated the efficient
logistical support provided by the staff of KEFJ, KATM and LACL. We
also thank the crew of the M/V Chigmit for providing logistical
support, safe passage, and expert guide service along the LACL
coast and crossing Cook Inlet. Field quarters in LACL were provided
by Silver Salmon Creek Lodge. Funding for the project was provided
by the National Park Service, Southwest Alaska Network Inventory
& Monitoring Program, the US Forest Service – Chugach National
Forest, and the Oil Spill Recovery Institute. Funding for the work
in Prince William Sound in 2006 was provided by the National
Science Foundation. Additional support was provided by the
University of Wyoming NSF-EPSCoR program, The College of Arts &
Sciences, and the Department of Zoology and Physiology. The Alaska
Department of Fish and Game provided logistical support and
personnel.
Literature Cited:
Almeida, F. C., L. S. Maroja, M. A. M. Moreira, H. N. Seuánez
and R. Cerqueira. 2005. Population structure and genetic
variability of mainland and insular populations of the Neotropical
water rat, Nectomys squamipes (Rodentia, Sigmodontinae). Genetics
and Molecular Biology 28: 693-699.
Amstrup, S. C., T. L. McDonald, and B. F. J. Manly, Eds. 2005.
Handbook of Capture-Recapture Analysis. Princeton University Press.
Princeton, New Jersey.
Bang, I. and C. N. K. Moores. 2003. The influence of several
factors controlling the interactions between Prince William Sound,
Alaska and the northern Gulf of Alaska. American Meteorological
Society 33: 19-36.
Beerli, P. and Felenstein, J. 2001. Maximum likelihood
estimation of a migration matrix and effective population sizes in
n subpopulations by using a coalescent approach. Proceedings of the
National Academy of Sciences of The United States of America 98:
4563-4568.
Beheler, A. S., J. A. Fike, L. M. Murfitt, O. E. Rhodes Jr., T.
S. Serfass. 2004. Development of polymorphic microsatellite loci
for North American river otters (Lontra canadensis) and
amplification in related Mustelids. Molecular Ecology Notes 4:
56-58.
39
-
Beheler, A. S., J. A. Fike, G. Dharmarajan, O. E. Jr. Rhodes,
and T. S. Serfass. 2005. Ten new polymorphic microsatellite loci
for North American river otters (Lontra canadensis) and their
utility in related mustelids. Molecular Ecology Notes 5:
602-604.
Ben-David, M., R. T. Bowyer, and J. B. Faro. 1996. Niche
separation by mink and river otters: coexistence in a marine
environment. Oikos 75: 41–48.
Ben-David M., G. M. Blundell, and J. E. Blake 2002. Post-release
survival of river otters: effects of exposure to crude oil and
captivity. Journal of Wildlife Management 66: 1208-1223.
Ben-David, M., G. M. Blundell, J. W. Kern, J. A. K. Maier, E. D.
Brown, and S. C. Jewett. 2005 Communication in coastal river
otters: creation of a variable resource shed for terrestrial
communities. Ecology 86: 1331-1345.
Blundell, G. M., M. Ben-David, P. Groves, R. T. Bowyer, and E.
Geffen. 2002. Sex-biased dispersal and gene flow in coastal river
otters: implications for natural recolonization of extirpated
populations. Molecular Ecology 11: 289–303.
Bowyer, R. T., J. W. Testa, J. B. Faro, C. C. Schwartz, and J.
B. Browning. 1994. Changes in diets of river otters in Prince
William Sound, Alaska: effects of the Exxon Valdez oil spill.
Canadian Journal of Zoology 72: 970–976.
Bowyer, R. T., J. W. Testa, and J. B. Faro. 1995. Habitat
selection and home ranges of river otters in a marine environment:
effects of the Exxon Valdez oil spill. Journal of Mammalogy 76:
1-11.
Bowyer, R. T., G. M. Blundell, M. Ben-David, S. C. Jewett, T. A.
Dean, and L. K. Duffy. 2003. Effects of the Exxon Valdez oil spill
on river otters: injury and recovery of a sentinel species.
Wildlife Monographs 153: 1–53.
Broquet, T. and E. Petit. 2004. Quantifying genotyping error in
noninvasive population genetics. Molecular Ecology 13:
3601-3608.
Burnham, K. P., and D. R. Anderson. 2002. Model Selection and
Multimodel Inference: A Practical Information-Theoretic Approach.
Springer, New York, New York, USA.
Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic
analysis: models and estimation procedures. Evolution 21:
550–570.
Crait, J. R., G. M. Blundell, K. E. Ott, J. K. Herreman, and M.
Ben-David. 2006. Late seasonal breeding of river otters in
Yellowstone National Park. American Midland Naturalist
156:189-192.
Dallas, J. F., and S. B. Piertney. 1998. Microsatellite primers
for the Eurasian otter. Molecular Ecology 7: 1247-1251.
Dallas J. F., K. E. Coxon, T. Sykes, P. R. F. Channin. 2003.
Similar estimates of population genetic composition and sex ratio
derived from carcasses and faeces of Eurasian otter Lutra lutra.
Molecular Ecology 12: 275-282.
DePue, J. E. and M. Ben-David. 2007. Hair snaring techniques for
river otters: a comparison of three trap designs. Journal of
Wildlife Management 71: 671-674.
Evanno, G., S. Regnaut, and J. Goudet. 2005. Detecting the
number of clusters of individuals using the software STRUCTURE: a
simulation study. Molecular Ecology 14:2611-2620.
Falush, D., M. Stephens, and J. K. Pritchard. 2003. Inference of
population structure using multilocus genotype data: Linked loci
and correlated allele frequencies. Genetics 164: 1567-1587.
40
-
Farrell, L. E., J. Roman, and M. E. Sunquist. 2000. Dietary
separation of sympatric carnivores identified by molecular analysis
of scats. Molecular Ecology 9: 1583–1590.
Felsenstein, J. 1989. PHYLIP - Phylogeny Inference Package
(Version 3.2). Cladistics 5: 164-166.
Foran, D. R., K. R. Crooks, and S. C. Minta. 1997. Species
identification from scat: an unambiguous genetic method. Wildlife
Society Bulletin 25: 835-839.
Frantz, A. C., L. C. Pope, P. J. Carpenter, T. J. Roper, G. J.
Wilson, R. J. Delahay, and T.Burke. 2003 Reliable microsatellite
genotyping of the Eurasian badger (Meles meles) using faecal DNA.
Molecular Ecology 12: 1649-1661.
Giblin, A. E., K. J. Nadelhoffer, G. R. Shaver, J. A. Laundre,
A. J. McKerrow. 1991 Biogeochemical diversity along a riverside
toposequence in Arctic Alaska. Ecological monographs 61:
415-435.
Golden, H. N., M. Ben-David, M. Seymour, and E. Solomon. 2009.
Estimating coastal river otter abundance and harvest potential in
the Kodiak Island Archipelago. Alaska Department of Fish and Game.
Progress Report. 15 Pp.
Goossens, B., L. Chikhi, S. S. Utami, J. de Ruiter, and M. W.
Bruford. 2000. A multi-samples, multi-extracts approach for
microsatellite analysis of faecal samples in an arboreal ape.
Conservation Genetics 1: 157-162.
Goudet, J. 2000. FSTAT. A program to estimate and test gene
diversities and fixation indices. Version 2.9.3.2.
www.unil.ch/lzea/softwares/fstat.htm.
Guertin et al. In review. Faecal genotyping reveals demographic
differences between subpopulations of river otters inhabiting a
contaminated coastal landscape. Molecular Ecology.
Guo S. W., and E. A. Thompson. 1992. Performing the exact test
of Hardy-Weinberg proportion for multiple alleles. Biometrics, 48:
361-372.
Hansen H. 2004. The effects of diet, extraction technique and
microsatellite primers on genotyping success and error rates of
river otter (Lontra canadensis) fecal DNA. MSc Thesis. University
of Wyoming, Laramie, WY.
Hansen, H., M. Ben-David, and D. B. McDonald. 2008. Effects of
genotyping protocols on success and errors in identifying
individual river otters (Lontra canadensis) from their faeces.
Molecular Ecology Resources 8: 282-289
Janssens, X., J. Michaux, M. C. Fontaine, R. Libois, J. de
Kermabon, P. Defourny, and P. V. Baret. 2007. Non-invasive genetics
at basin scale reveals barriers to the present otter recovery in
Southern France. Ecography: .
Kohn, M., F. Knauer, A. Stoffella, W. Schroder, and S. Paabo.
1995. Conservation genetics of the European brown bear – a study
using excremental PCR of nuclear and mitochondrial sequences.
Molecular Ecology 4: 95-103.
Larsen, D. N. 1983. Habitats, movements, and foods of river
otters in coastal southeastern Alaska. Thesis, University of Alaska
Fairbanks, Fairbanks, Alaska, USA.
Lebreton, J.-D., K. P. Burnham, J. Clobert, and D. R. Anderson.
1992. Modeling survival and testing biological hypotheses using
marked animals: A unified approach with case studies. Ecological
Monographs 62: 67-118.
41
-
Lilleskov, E. A., T. J. Fahey, T. R. Horton, G. M. Lovett. 2001.
Ectomycorrhizal fungal aboveground community change over an
atmospheric nitrogen deposition gradient. Ecological Applications
11: 397-410.
Lucchini, V., E. Fabbri, F. Marucco, S. Ricci, L. Boitani, and E
. Randi. 2002. Noninvasive molecular tracking of colonizing wolf
(Canis lupus) packs in the western Italian Alps. Molecular Ecology
11: 857 –868.
Melquist, W. E., and M. G. Hornocker. 1983. Ecology of river
otters in west central Idaho. Wildlife Monographs 83: 1-60.
Melquist, W. E., and A. E. Dronkert. 1987. River otter. Pages
626-641 in M. Novak, J. A. Baker, M. E. Obbard, and B. Malloch,
editors. Wild Furbearer Management and Conservation in North
America. Ministry of Natural Resources, Toronto.
Mills, L. S., J. J. Citta, K. P. Lair, M. K. Schwartz, and D. A.
Tallmon. 2000. Estimating animal abundance using noninvasive DNA
sampling: promise and pitfalls. Ecological Applications
10:283-294.
Morin, P.A., K. E. Chambers, C. Boesch, and L. Vigilant. 2001.
Quantitative polymerase chain reaction analysis of DNA from
noninvasive samples for accurate microsatellite genotyping of wild
chimpanzees (Pan troglodytes verus). Molecular Ecology 10:
1835-1844.
Murphy, M. A, L. P. Waits, K. C. Kendall. 2002. An evaluation of
long-term preservation methods for brown bear (Ursus arctos) faecal
DNA samples. Conservation Genetics 3: 435-440.
Nsubuga, A. M., M. M. Robbins, A. D. Roeder, P. A. Morin, C.
Boesch, and L. Vigilant. 2004. Factors affecting the amount of
genomic DNA extracted from ape faeces and the identification of an
improved sample storage method. Molecular Ecology 13:
2089-2094.
Ott, K. E., H. N. Golden, and M. Ben-David. (In prep a).
Recolonization or local reproduction? An assessment of river otter
recovery in previously-oiled areas of coastal Alaska via
non-invasive genetic sampling. Journal of mammalogy.
Ott K. E., S. A. Albeke, N. P. Nibbelink, and M. Ben-David. (In
prep b). Abundance and movements of coastal river otters:
implications for nutrient transports from sea to land. Journal of
Animal Ecology.
Paetkau, D. 2003. An empirical exploration of data quality in
DNA-based population inventories. Molecular Ecology 12:
1375-1387.
Pledger, S, K. H. Pollock, and J. L. Norris. 2003. Open
capture-recapture models with heterogeneity: I. Cormack-Jolly-Seber
model. Biometrics 59: 786-794.
Pollock, K. H. 2000. Capture-recapture models. Journal of the
American Statistical Association 95: 293-296.
P